HPE Is Also Having Trouble Making Money With AI Servers

My perspective as a does-anything dev who has gotten sucked into the ai sphere: The sort of ai server they sell is basically just large gpu farms, which require highly competent devs to actually utilize. The vast majority of "we use ai now" project integrations are farsical, and those that aren't are much better served by cloning an OpenAI large language model and custom training it (and even then you'll still end up with an LLM that can be ai 'jailbroken' just like the model it was trained on).